nov 1 - 2 2005: review meetingacclimate formation flying: a transition opportunity for blakchawks...

25
Nov 1 - 2 2005: Review Meeting ACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics and Intelligent Machines Laboratory University of California, Berkeley

Upload: kathlyn-stafford

Post on 19-Jan-2016

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Formation Flying: a Transition Opportunity for Blakchawks and Other

Rotorcrafts

Hoam Chung and Shankar SastryRobotics and Intelligent Machines Laboratory

University of California, Berkeley

Page 2: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Overview

Formation flight is the primary movement technique for helicopter teams

Formation flight automation can achieve:Diminishing aircrew stressMaintaining a formation under harsh battlefield conditionsOrganized management of

Formation joining/break upIn-flight formation changeScheduling of flight courseEmergency break up

Page 3: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Overview

The algorithm for an autonomous helicopter formation should be able to deal with

Maintaining a stable formation under disturbances

Mixed vehicle formations

Rejoin/breakup with guaranteed safety

Page 4: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Review: Mesh Stability

MotivationSmall disturbance can be amplified through the formation if information of neighboring vehicles is solely used

In order to allow large formations, this algorithm uses “leader information” as well as information from neighbors

Mesh StabilityDamping out error/disturbance propagation in the formation using leader information

Mesh stability is guaranteed for homogeneous formations

Page 5: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Review: Mesh Stability

Structure of mesh controller

Controller gainK(s)

Spacing errorw.r.t neighbors

Controller gainK(s)

Controller gainK(s)

Spacing errorw.r.t leader

+

+

Autonomously controlledunmanned helicopter

This achieves:

Page 6: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Review: Mesh Stability

Virtual leader running on a laptop

Real RUAVs

Virtual followers running on laptops

neighbor

neighbor

Page 7: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Review: Mesh Stability

Animation by A. Pant and X. Xiao

Page 8: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Review: Mesh Stability

e33

Gap errors are not damped out due to heterogeneityThe vehicle 33 is more agile than others

Using ‘leader information’ results in a dilemma

Page 9: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Review: Mesh Stability

The use of leader information improves the performance of the autonomous formation flight

For a heterogeneous mesh, an extension of mesh stability theory should be considered

The proposed mesh stability scheme has only one directional information flow

It is not easy to incorporate various objectives in mesh stability scheme

“Mesh Stability” does not mean the “Safety”

It’s a starting point for autonomous formation flight

Page 10: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Inputs from 160th SOAR

In many practical situations, a helicopter team is formed by heterogeneous vehicles

(ex. Little Birds + Blackhwaks + Chinooks)

In-flight manual formation joining process is extremely dangerous

Autonomous aerial refueling will be a great help

The effects of battlefield stress exerted on aircrew increases dramatically under tight formations and in adverse circumstances

Page 11: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Model Predictive Control

Computes control inputs using real-time optimization

Shows better performance than non-predictive controls

Can consider input/state constraints in on-line manner

Easily incorporates various control objectivesRelative gap maintenance

Tracking reference velocities and heading

Page 12: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Decentralized MPC

Gap error definitions: constant gap vs. varying gap

t

n

x yt

n

xy

Page 13: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Decentralized MPC

Finite-horizon optimal control (FHOC) problem

gap error tracking state input

Page 14: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Decentralized MPC

Simulation setupsRight-echelon 8-vehicle-formationInitial gaps are 30 ft. in x,y and zYamaha R-50 linear dynamics plus nonlinear kinematics model used

17 state variables, 4 inputsforward cruise condition, 30mi/h

DynOpt package is used for solving FHOC problemOnly current neighbors’ state variables are interchangedExtrapolate neighbors’ states for prediction

0

1

2

j

7

wind gust on vehicle 0

Page 15: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

0 5 10 15 20 25 30 35 40-2.5

-2

-1.5

-1

-0.5

0

0.5

Decentralized MPC

0 5 10 15 20 25 30 35 4014

16

18

20

22

24

26

28

30

32

gap

erro

rs in

x d

irect

ion

(ft)

Time (sec)

gap error 0-1gap error 1-2gap error 2-3gap error 3-4gap error 4-5gap error 5-6gap error 6-7

wind gust induced acceleration (ft/s^2)

0 -2

The maximum relative gap errors are damped

out successfully

Page 16: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Decentralized MPC

Scaling using Froude number (Mettler 2003)Ratio of inertia to gravitational forces

Dynamically similar if models have close Froude numbers

V: characteristic velocity (rotor tip speed)L: characteristic length (rotor radius)

Rotor Radius (ft) Rotor Speed (rad/s) Scale Froude Number

R22 13 53 0.38 1134

Virtual Model

10 62 0.5 1194

R-50 5 89 1 1230

Scale factor

Page 17: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Decentralized MPC

Vehicles 1,2,4,5,and 6 are replaced with scaled-up

virtual model

0 5 10 15 20 25 30 35 4014

16

18

20

22

24

26

28

30

32

gap

erro

r (f

t)

gap error 0-1gap error 1-2gap error 2-3gap error 3-4gap error 4-5gap error 5-6gap error 6-7

0

1

2

j

7

The maximum relative gap errors are damped out successfully also in the case of heterogeneous formation

Page 18: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Formation Manager

Autonomous formation should be reactiveDuring a formation flight, each vehicle may faces various situations

A vehicle in a formation have multiple modes

A high level agent on top of MPC can make autonomous formation safer and more flexible

Formation ManagerScheduling normal breakup/rejoin

Managing course changes and in-flight reconfiguration

Dealing with emergency situations

Managing communication channels

Page 19: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Formation Manager

Overall system structure

FormationManager

MPCController

controlinput

Helicopter

ReferencesRelative gaps

OperatorCommands

Navigation infofrom neighbors

Modified Reference

Vehicle states

Mode

Page 20: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Formation Manager

Inside of the formation manager

Modified Reference

Vehicle states

Finite State Machine

Operator command

s

TrajectoryInterpolator

Navigation infofrom

neighbors

WaypointsReference velocity/headingvalues

MPCTrajectoryGenerator

Mode

Page 21: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Formation Manager

Simple FSM for emergency break up/rejoin

Normal

NormalAs the last follower

FormationAway from the formation

Normal

Single

Approach to the formation

Escape

Sufficientspacing

Rejoiningrequested

Gap errorsmall enough

Page 22: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Formation Manager

Simulations

a. Normal formation b. Proximity warning c. Escape from the formation

d. Fly away from the formation e. Ready to approach f. Rejoin the formation

Page 23: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Conclusions

Without any explicit disturbance rejection algorithm, the proposed MPC based formation shows mesh stable property even for heterogeneous formation

MPC with the proposed formation manager can deal with various formations and many practical issues

Break up/rejoinIn-flight reconfiguration

For safer and more versatile autonomous formation, a formation manager should be implemented as a high level agent

Page 24: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Future Works

Currently, this research is supported by Phase I of ARO STTR with Scientific Systems Company, Inc.

In Phase II, the proposed MPC-based formation flight will be implemented on our BEAR fleet and a series of flight experiments will be performed

Ursa Major 1 Ursa Magna 1

Page 25: Nov 1 - 2 2005: Review MeetingACCLIMATE Formation Flying: a Transition Opportunity for Blakchawks and Other Rotorcrafts Hoam Chung and Shankar Sastry Robotics

Nov 1-2 2005: Review Meeting ACCLIMATE

Future Works

The formation manager will be modeled as a hybrid system using hybrid CAD tools like HyVisual

(Part of) formation manager functionality will be implemented on-board and in-flight reconfiguration will be demonstrated

Technology transition to Blackhawk:

Human operator interaction is another future research subject